Analytic Inaccuracy Resulting from Hematology Specimen Characteristics

Transcription

1 Analytic Inaccuracy Resulting from Hematology Specimen Characteristics Three Cases of Clinically Misleading Artifacts Affecting White Blood Cell and Platelet Counts Sporadically, hematology analyzers generate grossly erroneous results for one or several parameters of the complete blood count (CBC) because of a characteristic or peculiarity of an individual patient specimen. This article presents three cases of relatively less well-known analytic errors of this sort, falsely elevating white blood cell count and platelet count and decreasing total white blood cell count, as determined on a Coulter Aperture Impedence Analyzer. The type of alteration in the CBC parameters and the instrument histogram abnormalities seen in each of these different types of analytic inaccuracy are discussed. (Key words: Analytic artifact; Analytic error; White blood cell count; Platelet count; Automated hematology) Am J Clin Pathol 1989;92: Received June 23, 1988; received revised manuscript and accepted for publication March 6, Address reprint requests to Dr. Savage: Department of Pathology, Millard Fillmore Hospitals, 3 Gates Circle, Buffalo, New York RICHARD A. SAVAGE, M.D. Department of Pathology, Millard Fillmore Hospitals, Buffalo, New York AUTOMATED HEMATOLOGY ANALYZERS have been found to be prone to a variety of analytic errors caused by characteristics of individual patient specimens. These specimen-analyzer incompatabilities produce erroneous data on individual patients, a form of analytic inaccuracy not detectable by routine calibration or longitudinal process control techniques. Some analytic errors are universal to all analyzers, such as counting nucleated red blood cells (RBCs) as leukocytes. Some involve only one system with certain analytic principles but not others. For example, lipemia will elevate the hemoglobin on Coulter -type aperture impedance counters 10 but lower the hemoglobin value on an Ortho Laser Analyzer. 14 One especially vexing problem is analytic artifacts spuriously altering white blood cell (WBC) count or platelet count. The RBC parameters measured by hematology analyzers are interdependent; most analytic artifacts will produce an alteration in the expected relationships between hemoglobin, hematocrit, and RBC count, 13 which will serve as a flag for detection of the error. No comparable safeguard interdependence exists for the WBC count or platelet count. The WBC count may be spuriously el- evated (pseudoleukocytosis), most often because of small particles or debris, such as platelet clumps, 5 ' 6 " 12 apoproteins, 3 ' 4,7 or unlysed RBCs. 1 Much less common is a spuriously decreased WBC count (pseudoleukopenia) or disproportionate decreases in one cell class. Pseudoneutropenia resulting from leukoagglutinins has been reported in cirrhotic patients with liver failure, 2 and balanced leukopenia of approximately X 10 9 /L ( X 10 3 / fil) was reported in uremic and immunosuppressed patients whose blood was analyzed on Coulter instruments. 8 Both of these problems are relatively rare. Spurious thrombocytopenia resulting from platelet clumping is a real but uncommon problem, occurring in 0.1% of complete blood counts (CBCs). 12 Detection and validation procedures for spurious thrombocytopenia have been published. 612 More rarely, the platelet count may be spuriously elevated, most commonly because of cryoproteins 1,7 or cytoplasmic microfragments derived from degenerating WBCs or RBCs. 1 ' 715 If the platelet count is not below the lower limit of normal, the likelihood of detection of this analytic problem is small unless RBC parameters are affected or a differential count and slide review happens to be performed on the affected blood. In the past few years, newer hematology analyzers have been developed that can display histograms illustrating the size distribution of the WBCs, RBCs, and platelet populations in each patient. These instruments also have a variety of internal flagging routines to detect WBCassociated analytic errors based on the slope of the curve, the shape of the peaks, and the presence of valleys separating populations of cells of different sizes. The use of such newer analyzers has led to the discovery of some analytic errors and has made detection of ones previously known much easier. The purpose of this communication is to present three cases illustrating the use of CBC and 295

2 296 SAVAGE AJ.C.P. September 1989 Table 1. CBC Results from Patient 1, with Small Pseudoparticles Simulating Leukocytes Coulter STKR Coulter S770 REL ND SUSPECT; EOS IMM GRANS ATYP LYMPHS BLASTS WBC, X107MX10 3 //il) RBC, X10 12 /L(X10 6 /ML) MCHC, g/l (g/dl) 24* (9.2) (23.9) (38.6) 14.9f (9.1) (24.7) (36.9) SI units (conventional units). ' Count backlit (flagged for review) by instrument. t No flags for erroneous results produced (actual WBC count 1.2 X I0 9 /L) (XlOVfiL). Platelet count could not be derived by the instrument. histogram data from a Coulter STKR and S+IV Analyzer to detect patients with falsely elevated WBC counts and platelet counts and spuriously decreased total WBCs. Case 1: Falsely Elevated WBC Counts Resulting from Large Particulates (approximately RBC size) This patient was a 61-year-old woman with pulmonary adenocarcinoma involving thoracic lymph nodes. After surgical resection, she was treated with methotrexate 50 mg, intravenously; Cytoxan 800 mg, intravenously; and oral CCNU. A prechemotherapy blood count gave the results shown in Table 1. The histograms from this blood count are shown in Figure 1. Previous and subsequent blood counts showed no such analytic artifacts. The actual WBC count on this specimen, performed in duplicate in a chamber, was 1.2 X 10 9 /L (1.2 X 10 3 /ML). At no point could the instrument generate a platelet count. The blood was also run in a satellite facility, with the use of a Coulter S770, which had been previously "calibrated" to yield values similar to those from the STK-R used for the initial analyses. These results are seen in Table 1. Case 2: Pseudonucleated RBCs and Pseudoplatelets Resulting from Small Particulate Material The patient was a 31-year-old man suspected of having early human immunodeficiency virus (HIV) infection. His initial CBC results are listed as "Count 1" in Table 2. The specimen was warmed at 37 C and reanalyzed under the supposition that the high MCHC initially observed resulted from the presence of cold agglutinins. No significant alteration in values was produced by this maneuver, the results after prewarming are listed as "Count 2" in Table 2. No histogram was printed for these results. The problem was left overnight to be resolved on day shift. The following morning the initial blood was reanalyzed with the result seen W3C RBC _L 400 AM I SO SUSPECT= MRBC ABNORMAL DISTRIBUTION 100 _L l 200 SUSPECT^ PLT CLUMPS 100" FlG. 1. Coulter histograms from patient 1. Nonlysable 35-fL particulates are seen on WBC histogram (solid arrow) and RBC histogram (open arrow). as "Count 3" in Table 2. The RBC count had decreased slightly, and small nonlysable particles simulating nucleated RBCs or microcytes were present, as seen in Figure 2. Similar particles produced a grossly distorted platelet histogram; nonetheless, the analyzer was able to produce a "platelet count" withoutflagging.the patient's blood was redrawn and reanalyzed; these results are listed in the fourth column of Table 2. The histograms for this CBC were entirely normal. Smear examination showed no evidence of nucleated RBCs, giant platelets, or platelet clumps. There were no clots in the specimen. Case 3: Histograms in a Patient with WBC Agglutinates, Producing Spuriously Decreased WBC Count This patient was a 35-year-old white woman with a long-standing case of alcohol abuse and alcoholic cirrhosis. She was hospitalized with fever, later proved to be Legionella pneumonia. Her CBC values are given in Table 3. Her actual WBC count was 7.2 X 10'/L (7.2 X 10 3 /ML), with a left shift. Well-formed leukoagglutinates found in the feathered edge of the smear included all WBC classes (Figs. AA and B). The differential performed on the isolated cells in the usual working area of the smear was normal, and the automated CBC results generated by the Coulter were similar to the isolated cell differential (Table 4). The analyzer's Rl/ R2flagsand histograms indicate the likelihood of an erroneous WBC result (Fig. 3). Table 2. Progressive Development of Pseudonucleated Red Blood Cells and Pseudoplatelets in Case 2* Count 1 Time 0 Count 2 One Hour Later Count 3 Following A.M. New Specimen, Next Day WBC, X10 9 /L(X10 3 /ML) RBC, X10 12 /L(X10 6 /ML) MCHC, g/l (g/dl) Pit, X107L(X10 3 /^L) (15.2) (39.6) (38.4) (15.0) (37.4) (40.1) (15.2) 0.36 (36) (42.2) (15.1) (35) Coulter STKR* instrument; effect was observed with the Coulter S + IV* also.

3 Vol. 92 No. 3 ARTIFACTS AND CELL COUNTING 297 GRANULOPENIA SUSPECT^ EOS MM GRANS ATY? LYflPHS BLASTS FIG. 2. Coulter histograms from patient 2. Nonlysable particles are seen on RBC and WBC histograms (solid arrows) plus small particles on platelet count histogram (open arrow). The platelet count curve does not return to baseline at 20 fl (curved arrow), indicating inaccuracy of instrument platelet count. Discussion Cases 1 and 2 illustrate interference by optically invisible pseudoparticles producing spurious increases in WBC count in case 1 and in platelets in case 2. The particle seen in the WBC histogram in case 1 produced tracings similar to those reported for cryoglobulin interference 3 ; a similar "ski-jump" pattern can be produced by the lipids given for total parenteral nutrition therapy (Ward PCJ. Personal communication) and by Heinz bodies and the globin chain inclusions in thalassemia. 16 None of these possible causes for the WBC anomaly was present in case 1; the most likely explanation for the analytic error is interference by some component of the chemotherapy mixture unique to this protocol. The analytic error seen in case 1 also occurred in a second patient at the Millard Fillmore Hospital (data not presented but similar to case 1). Her WBC count was less than 0.1 X 10 9 /L (0.1 X 10 3 / ML), but the S+IV count was 1.2 X 10 9 /L. (1.2 X 10 3 / M L). This patient also had lung carcinoma and was being Table 3. CBC Values in Patient 3, a Patient with Leukocytic Autoagglutination WBC, X10'/L(X10 3 A»L) RBC, XlO'VLfXIOV/iL) HCHC, g/l (g/dl) Pit, X10 9 /L(X10 3 AJL) * WBC count backlit (flagged for review). 4.3* (10.6) (29.6) (35.3) 70 Table 4. Comparison of Manual Slide Differential and Coulter Three Part Differential Results on Case 3 Segs, % Bands Eosinophils Lymphocytes, % Monocytes, % Manual Diff. 6 8 ^ ^>84 16-"^ Granulocytes, % (RI flags present) Lymphocytes, % Mononuclear cells, % Automated Diff treated with the same chemotherapeutic protocol as was patient 1. Because many chemotherapy units use smaller hematology analyzers without sophisticated histographic evaluation, similar to the S770, clinicians and laboratorians should be alert for the chance occurrence of falsely elevated WBC counts in patients having chemotherapy and be prepared to validate counts against differential smears. Recognition of the analytic error in case 1 was also possible because of alterations in RBC parameters, such as the MCHC of 386 g/l (38.6 g/dl) and the hematocrit/hemoglobin ratio of 2.6:1. Although the presence of small particles did produce a disturbance in expected interdependent RBC parameter relationships, these perturbations do not suggest a primary analytic defect affecting the WBC count. Accordingly, validation against smears still appears to be the best course. Case 2 illustrates a slightly different analytic artifact. The particles appear to be smaller than those that were present in case 1, with a significant number in the 2-5- fl range on the platelet histogram (Fig. 2). A spun microhematocrit on the case 2 blood showed a thin layer of m WBC PBC LYMPHOPENIA _L 0 130" SUSPECT' PLT CLUMPS 0PMAL DISTRIBUIION ABNC - PEjlfOLlVEftd FIG. 3. Coulter histograms from patient 3. WBC Rl (open arrow) and R2 (closed arrow) histogram valleys are absent, indicating inaccuracy of WBC differential data. WBC count is backlit (flagged), indicating inaccuracy of WBC count.

4 298 SAVAGE A.J.C.P. September 1989 FIG. 4. A (left) and B (right). Leukoagglutinates from case 3. Lymphocytes (open arrows) and monocytes (solid arrows) are present in agglutinates in addition to neutrophils. Wright (X800). translucent liquid atop the plasma. Unstained air-dried smears from this layer showed small colorless droplets, smaller than an RBC and similar to the smaller particles seen on the platelet histogram. The origin of this material is a mystery, but its effect seems to have been to form small droplets simulating platelets and small RBCs. The progressive nature of the occurrence of this phenomenon is shown in Table 2. The presence of these small particles was seen easily on the histograms, and the process produced a high MCHC and a hematocrit to hemoglobin ratio of 2.6:1. These alterations are most likely sufficient to prompt reanalysis and reevaluation of the specimen on an analyzer not equipped with histograms. The effect on the WBC count in case 2 appears to have been small (Table 2). A final point of interest was that the analyzer was able to compute a platelet count, although the result was grossly in error and the histogram shows major abnormalities. The STKR flagged this platelet count for review; older models, such as the S+II, do not have flags of this type and the operator must be alert for aberrant histogram values such as the one illustrated in case 2 in order to avoid releasing erroneous platelet counts. Patient 3 is similar to the two patients reported by Epstein and Kruskal. 2 In the present patient, however, the aggregates involved all WBC classes (Figs. 4A and B), rather than just neutrophils, and the effect produced was balanced leukopenia rather than prominent neutropenia. The Coulter flow-through "three-part" differential closely matched the manual slide differential performed on the isolated cells, as seen in Table 4. The S+IV used to analyze case 3 produced histogram warnings and flagged (backlit) the WBC count results, indicating the probability of a WBC/differential analytic error. The older S+ used in the Epstein study did not have these error-detection features, and, accordingly, inappropriate diagnostic investigation of the pseudoneutropenia seen in these patients ensued. These three cases illustrate several points regarding the validity of results from Coulter-type aperture impedence automated hematology analyzers. First, despite instrument-flagging codes, parameter backlighting, and valley alarms, simple review of the histograms by the instrument operator provides valuable clues to the presence and type of analytic errors. An abnormality in the log-normal shape of the platelet histogram, even if a curve is fitted and a result generated, suggests inaccuracy of the platelet result. The presence of an altered hematocrit-hemoglobin ratio and a nonphysiologic MCHC should trigger review of WBC and platelet result accuracy, as well as validation of RBC parameters. Finally, operators of older aperture

5 vol.92-no.3 ARTIFACTS AND CELL COUNTING 299 impedence analyzers lacking histogram displays must be even more alert to the presence of spurious, specimenrelated analytic inaccuracy. Only by carefully following error-detection and validation routines can the hematology laboratory avoid issuing clinically misleading results when random, specimen-related analytic error is present. References 1. Cornbleet P. Spurious results from automated hematology analyzers. Laboratory Medicine 1983;14: Epstein HD, Kruskall MS. Spurious leukopenia due to in vitro granulocytic aggregation. Am J Clin Pathol 1988;89: Fishleder AJ, Hoffman GC. Automated hematology: counts and indices. Laboratory Management 1984; Guliani GL, Hyun BH, Gabaldon H. Falsely elevated automated leukocyte count on cryoglobulinemic and/or cryofibrinoginemic samples. Laboratory Medicine 1977;8: Henderson SJ, Wood JK. Factors producing error in the total leukocyte count as measured on the Coulter S+IVD. Clin Lab Haematol 1986;8: Lombarts AJPF, de Kieviet W. Recognition and prevention ofpseudothrombocytopenia and concomitant pseudoleukocytosis. Am J Clin Pathol 1988;89: Lucas FV, Miller ML, Savage RA, Fishleder AJ. Cryoglobulinemia. Check Sample H84-1 (H-144). Chicago, IL: American Society of Clinical Pathologists, Luke RG, Koepke JA, Siegel RR. The effects of immunosuppressive drugs and uremia on automated leukocyte count. Am J Clin Pathol 1971;56: Malcom ID, Monks P, Katz M. Spurious thrombocytosis in acute myelocytic leukemia. N Engl J Med 1978;298: Nosanchuk JS, Roark MF, Wanser C. Anemia masked by triglyceridemia. Am J Clin Pathol 1974;62: Payne BA, Pierre RV. Pseudothrombocytopenia: a laboratory artefact with potentially serious consequences. Mayo Clin Proc 1984;59: Savage RA. Pseudoleukocytosis due to EDTA induced platelet clumping. Am J Clin Pathol 1984;81: Savage RA. Ask the experts: inconsistent relationship between hemoglobin and hematocrit. Laboratory Medicine 1987; 18: Savage RA. Lipemia and the Ortho ELT-8. Response to letter of Martha Sheldon MT (ASCP). Summing Up (CAP Today), Spring 1988; Stass SA, Holloway ML, Peterson V, Creegan WJ, Gallivan M, Schumacher HR. Cytoplasmic fragments causing spurious platelet counts in the leukemic phase of poorly differentiated lymphocytic lymphoma. Am J Clin Pathol 1979;71: Sultan C. Discussion of case 7. In: Proceedings of the Coulter Automated Differential International Symposium. P/N A. Hialeah, FL: Coulter Electronics, 1987;21.